5422
Biochemistry 2001, 40, 5422-5432
Involvement of the Arg-Asp-His Catalytic Triad in Enzymatic Cleavage of the Phosphodiester Bond† Robert J. Kubiak,‡,§ Xiangjun Yue,‡ Robert J. Hondal,|,⊥ Cornelia Mihai,‡ Ming-Daw Tsai,| and Karol S. Bruzik*,‡ Department of Medicinal Chemistry and Pharmacognosy, UniVersity of Illinois at Chicago, Chicago, Illinois 60612, and Departments of Chemistry and Biochemistry, The Ohio State UniVersity, Columbus, Ohio 43210 ReceiVed October 12, 2000; ReVised Manuscript ReceiVed March 15, 2001
ABSTRACT: Phosphatidylinositol-specific phospholipase C (PI-PLC) catalyzes the cleavage of the P-O bond in phosphatidylinositol via intramolecular nucleophilic attack of the 2-hydroxyl group of inositol on the phosphorus atom. Our earlier stereochemical and site-directed mutagenesis studies indicated that this reaction proceeds by a mechanism similar to that of RNase A, and that the catalytic site of PI-PLC consists of three major components analogous to those observed in RNase A, the His32 general base, the His82 general acid, and Arg69 acting as a phosphate-activating residue. In addition, His32 is associated with Asp274 in forming a catalytic triad with inositol 2-hydroxyl, and His82 is associated with Asp33 in forming a catalytic diad. The focus of this work is to provide a global view of the mechanism, assess cooperation between various catalytic residues, and determine the origin of enzyme activation by the hydrophobic leaving group. To this end, we have investigated kinetic properties of Arg69, Asp33, and His82 mutants with phosphorothioate substrate analogues which feature leaving groups of varying hydrophobicity and pKa. Our results indicate that interaction of the nonbridging pro-S oxygen atom of the phosphate group with Arg69 is strongly affected by Asp33, and to a smaller extent by His82. This result in conjunction with those obtained earlier can be rationalized in terms of a novel, dual-function triad comprised of Arg69, Asp33, and His82 residues. The function of this triad is to both activate the phosphate group toward the nucleophilic attack and to protonate the leaving group. In addition, Asp33 and His82 mutants displayed much smaller degrees of activation by the fatty acid-containing leaving group as compared to the wildtype (WT) enzyme, and the level of activation was significantly reduced for substrates featuring the leaving group with low pKa values. These results strongly suggest that the assembly of the above three residues into the fully catalytically competent triad is controlled by the hydrophobic interactions of the enzyme with the substrate leaving group.
Phosphatidylinositol-specific phospholipase C (PI-PLC)1 is a key effector of signal transduction in higher organisms (1). This enzyme is activated in response to binding of a hormone to an extracellular receptor and cleaves the P-O bond of phosphatidylinositol (PI, 1) by a mechanism that involves the 2-hydroxyl group of inositol as an intramolecular nucleophile. In the past, we have determined that the mechanism of enzymes from mammalian tissues at the substrate level is analogous to those of the simpler bacterial enzymes (2-4). The bacterial enzymes differ from the mammalian proteins in that the intramolecular nucleophilic attack of the 2-hydroxyl group on the phosphorus atom (Scheme 1) to form inositol 1,2-cyclic phosphate (2, IcP) is much faster than the subsequent hydrolysis of IcP to inositol † This work was supported in part by NIH Grants GM30327 (M.D.T.) and GM57568 (K.S.B.). The NMR spectrometer was supported by NIH Grant S10 RR11283. * To whom correspondence should be addressed. Telephone: (312) 996-4576. Fax: (312) 996-7107. E-mail:
[email protected]. ‡ University of Illinois at Chicago. § Present address: Department of Chemistry, The Scripps Research Institute, 10550 N. Torrey Pines Rd., La Jolla, CA 92037. | The Ohio State University. ⊥ Present address: Department of Biochemistry, University of WisconsinsMadison, Madison, WI 53706.
Scheme 1
1-phosphate (3, IP) (5, 54), while the mammalian enzymes hydrolyze the cyclic phosphate prior to releasing it from the active site (3, 6). Despite this difference, the more stable and smaller bacterial enzymes can be used as surrogate models of mammalian PI-PLC for studying chemistry that is underlying enzymatic activity. PI-PLC belongs to a broader class of phosphodiesterases that cleave β-hydroxyalkyl phosphoesters by the attack of an internal hydroxyl group on phosphorus to form a fivemembered cyclic phosphate as an intermediate. This category of enzymes includes a large family of RNases among others (7); it is therefore not surprising that many analogies exist between mechanisms of RNases and PI-PLC, despite their vastly different structures. According to the most widely accepted mechanism of RNase A (7-9), the 2′-hydroxyl
10.1021/bi002371y CCC: $20.00 © 2001 American Chemical Society Published on Web 04/14/2001
Novel Mechanism of Phosphodiester Cleavage
FIGURE 1: Comparison of major components of the active sites of PI-PLC and ribonuclease A.
group is deprotonated by the His12 residue (general base) and attacks the phosphorus atom (Figure 1A), while the 5′oxygen is protonated by the imidazolium form of His119 (general acid). The function assigned to Lys41 is the subject of controversy (10), and ranges from forming a strong, lowbarrier hydrogen bond to the nonbridging oxygen of the phosphate group (8, 11) to electrostatic stabilization of the negative charge in the transition state (12). The X-ray structure of Bacillus cereus PI-PLC (13, 14) showed that, as for RNase A, its active site contains two histidines, His32 and His82, as well as Arg69 (Figure 1B). There is a major distinction between the two active sites, however, in that both histidines in PI-PLC are associated with aspartate residues to form His32-Asp274 and His82Asp33 diads (15, 16). In addition, the His32-Asp274 diad further interacts with the 2-hydroxyl group of inositol (14), to form a catalytic triad analogous to those found in serine proteases. The X-ray structure of the complex of PI-PLC with a partial substrate analogue, inositol (13, 14, 17), showed that Asp274 and His32, and His32 OH2, are within hydrogen bonding distances of 2.7 and 2.8 Å, respectively. In addition, the carboxyl function of Asp274 and the imidazole residue of His32 are coplanar (14), in optimal arrangement for charge relay within the triad. Furthermore, a low-field NMR signal observed in 1H NMR spectra of PIPLC suggested the existence of a low-barrier hydrogen bond between His32 and Asp274 (18). Thus, the structure and the catalytic function of the nucleophile-activating part of the active site seem well optimized for catalysis, both in the unliganded enzyme and in its complex with inositol. The subsequent studies by site-directed mutagenesis fully confirmed the functional significance of His32 and Asp274 as that of a complex general base (15). 1 Abbreviations: diC PC, 1,2-dihexanoyl-sn-glycero-3-phosphocho6 line; DIPEA, diisopropylethylamine; DOsPI, (2R)-1,2-dioctanoyloxypropanethio-3-(1-phospho-1D-myo-inositol); DOsPsI, (2R)-1,2-dioctanoyloxypropanethio-3-(1-thiophospho-1D-myo-inositol); DPPI, 1,2dipalmitoyl-sn-glycero-3-(1-phospho-1D-myo-inositol); DPPsI, 1,2dipalmitoyl-sn-glycero-3-(1-thiophospho-1D-myo-inositol); DPsPI, (2R)1,2-dipalmitoyloxypropanethio-3-(1-phospho-1D-myo-inositol); DTP, 4, 4′-dithiobispyridine; ES-MS, electrospray mass spectrometry; Glu1-P, glucose 1-phosphate; Gro-PI, sn-glycero-3-(1-phospho-1D-myoinositol); Gro-sPI, (2R)-1,2-dihydroxypropanethio-3-(1-phospho-1Dmyo-inositol); HDPC, hexadecylphosphorylcholine; HPAEC, high-pH anion exchange chromatography; IP, myo-inositol 1-phosphate; IcP, myo-inositol 1,2-cyclic phosphate; IcPs, myo-inositol 1,2-cyclic phosphorothioate; MOPS, morpholinepropanesulfonic acid; NPIPs, pnitrophenyl 1D-myo-inositol phosphorothioate; PAD, pulsed amperometric detection; PI, phosphatidylinositol; PI-PLC, phosphatidylinositolspecific phospholipase C; SDC, sodium deoxycholate; SDM, sitedirected mutagenesis; TBP, trigonal bipyramidal; UcP, uridine 2′,3′cyclicphosphate;3′-UMP,uridine3′-phosphate;3′,5′-UpA,uridyl(3′,5′)uridine; WT, wild type.
Biochemistry, Vol. 40, No. 18, 2001 5423 In contrast, the available X-ray structures provide much less information about two other elements of mechanism: general acid catalysis and phosphate activation. These are also aspects of catalysis which are subject to perennial debate in the case of ribonucleases. Our previous SDM study indicated that His82 is involved in interactions with Asp33 to form a general acid diad (shown in Figure 10 of ref 15), the function of which is to assist the leaving group by protonation or hydrogen bonding. This conclusion has been reached on the basis of kinetic evaluation of His82 and Asp33 mutants with substrates featuring low-pKa leaving groups, such as p-nitrophenoxide (15) or mercaptide (15, 19). In general, Asp33 and His82 mutants displayed much higher reactivity with these substrates than would have been predicted from the reactivity of these mutants with the normal substrate. These results are supported by the X-ray structure of the PI-PLC-inositol complex (13) which indicates hydrogen bonding interaction between His82 and Asp33. The relatively large distance between these residues (3.2 Å) and the perpendicular orientation of the carboxylate and imidazole groups indicate, however, that this diad is not in the optimal arrangement to perform the charge relay function. Another major disparity between the RNase A and PIPLC is in the mode of stabilization of the negative charge of the phosphate group. This is manifested by the large difference in the magnitude of kinetic effects of sulfur substitution for the nonbridging phosphate oxygens (kO/kS, thio effect) in substrates for both enzymes. For example, the RNase A-catalyzed cleavage of Rp- and Sp-diastereomers of 2′,3′-cyclic nucleoside phosphorothioate and 3′,5′-dinucleoside phosphorothioate showed a small thio effect ranging from 2 to 70 (10, 20, 21). In contrast, cleavage of phosphorothioate analogues of IcP (IcPs) and dipalmitoylphosphatidylinositol (DPPsI) by bacterial PI-PLC displayed an extremely high thio effect (kO/kS ) 1.6 × 105) (19, 22). The facts that substitution of the pro-S oxygen with sulfur results in resistance to cleavage and that mutation of Arg69 to lysine reduces the Sp-thio effect by four orders (!) of magnitude (22) suggest that Arg69 is involved in interaction with the pro-S oxygen atom. The possibility that the interaction of Arg69 with the pro-S oxygen is quite feasible was confirmed by molecular modeling of the complex of PI-PLC with O-methyl inositol 1,2-cyclic phosphorane (pentacoordinated analogue of the TBP transition state) which showed that Arg69 is within 2.8 Å of the pro-S nonbridging oxygen of the phosphorane, while His82 is 3.7 Å away from the glycerol sn-3 oxygen (K. S. Bruzik and P.-G. Nyholm, unpublished results). The structural data acquired previously did not indicate any functional relationship between Arg69 and the general acid diad. Although the X-ray structure showed proximity of Arg69 and Asp33 (3.0 Å) (14), the functional significance of this interaction was unclear. In the past, we have described structure-function studies (15, 16, 19, 22-24) aimed at determining the catalytic roles of individual amino acid residues of the active site. The focus of this paper is to describe a global view of the catalytic site, and emphasize coordination of the roles of several catalytic functions into a single catalytic machinery. To achieve this goal, we have employed the concept of “matched substrate-enzyme mutagenesis”. In this approach, we compare kinetic parameters of four pairs of reactants: (i) WT enzyme and natural substrate, (ii) mutant enzyme and natural
5424 Biochemistry, Vol. 40, No. 18, 2001
Kubiak et al. MATERIALS AND METHODS Materials
FIGURE 2: Structures of substrate analogues used in this work.
substrate, (iii) WT enzyme and substrate analogue, and (iv) mutant enzyme and substrate analogue. In the last pair, alteration at the specific substrate site is matched by mutation of a specific enzyme residue suspected of performing functional interaction with that substrate site. If the decreased catalytic rates in the cases of pairs ii and iii as compared to that of pair i are due to the removal of the same stabilizing factor, no further activity decrease should be observed in case iv. This methodology enables detailed description of the functional relationship between the substrate and enzyme residues and further refinement of the enzyme mechanism. To investigate interactions between the phosphate and active site residues, we have used the natural substrate DPPI (4, Figure 2), its phosphorothioate analogues (Rp)- and (Sp)DPPsI (5 and 6, respectively), phosphorothiolate analogues DOsPI and DPsPI (7 and 8, respectively), and phosphorodithioate analogues (Rp)- and (Sp)-DOsPsI (9 and 10, respectively). Our results support strong cooperative effects between substrate and enzyme residues and in particular indicate involvement of a novel catalytic triad, including His82, Asp33, and Arg69, performing dual tasks of leaving group protonation and phosphate activation (23). In addition, we employed the same approach to study the effect of the full structure of the leaving group on the catalytic mechanism. In this case, we used substrate analogues devoid of hydrophobic chains and featuring leaving groups with varying pKa values, such as sn-glycero-3-(1-phosphoinositol) (Gro-PI, 11) 1,2-dihydroxypropanethio-3-(1-phospho-1D-myo-inositol) (GrosPI, 12), and the corresponding p-nitrophenyl thiophosphoinositols, (Sp)-NPIPs (13) and (Rp)-NPIPs (14). The results obtained from the latter series of experiments present clear evidence that the assembly of the Arg, Asp, and His residues into the functional catalytic triad is affected by hydrophobic interactions of the enzyme with hydrocarbon chains of the substrate leaving group.
DPPI and DPsPI were obtained analogously as described recently (25, 26) starting from 2,3,4,5,6-pentakis(O-methoxymethylene)-myo-inositol (15; see the Supporting Information), and DPPsI was synthesized as reported previously (3). Natural PI was purified from soy bean phosphoinositide (Sigma) by chromatography on silica gel using a chloroform/ methanol/water/ammonium hydroxide mixture (1.6:0.8:0.05: 0.01, v/v) as the eluting solvent. NPIPs was synthesized as reported recently (15), and NPIP (27) was prepared by a slight modification of this procedure. Gro-PI (28) was obtained by deacylation of DPPI with methylamine in a methanol solution at room temperature over the course of 24 h. The fatty acid methylamide was removed by extraction of the aqueous solution of the reaction mixture with a methanol/chloroform mixture (4:6), and the aqueous phase was concentrated to give the pure product. Due to competing cyclization, DPsPI was deacylated into Gro-sPI using the lipase-catalyzed reaction (see the Supporting Information). Diisopropylethylamine (DIPEA), O-methyl phosphorodichloridite, 4,4′-dithiopyridine (DTP), and tetra-n-butylammonium periodate were from Aldrich. Esterase, MOPS, and Trizma base were from Sigma. All organic solvents were from Fisher, and were not further purified for chromatographic use. For their use as reaction media, solvents were dried over appropriate desiccants, stored in vacuum ampules prior to use, and transferred directly to reaction glassware by distillation under vacuum. WT PI-PLC and its mutants were expressed and purified as reported previously (15, 22). All NMR spectra were recorded with a Bruker DPX-300 spectrometer. The ES-MS spectra were obtained with a Micromass Quatro II tandem mass spectrometer. Assays of PI-PLC ActiVity 31P NMR Assay. All kinetic runs were performed in 50 mM MOPS buffer (sodium salt) at pH 7.0 in the presence of 5 mM Na-EDTA and 10% D2O. The substrate concentration was 10 mM, and the detergent concentration was 40 mM, except for sodium deoxycholate which was used at a concentration of 20 mM. Glucose 1-phosphate (Glu-1-P) was used as an internal standard for quantitation. Each sample was prepared in the following manner. The appropriate amounts of the substrate and detergent in an Eppendorf tube were dispersed by vortexing in the buffer solution (200 µL) containing 100 mM MOPS, 10 mM EDTA (pH 7.0), H2O (156 µL), D2O (40 µL), and a glucose 1-phosphate solution (0.56 M, 4 µL) and quantitatively transferred into a 5 mm NMR tube. Prior to the assay, the sample was sonicated in the ultrasonic bath to achieve optical transparency. Kinetic runs were performed at 25 °C, and spectra were recorded at 121.5 MHz with a Bruker DPX-300 spectrometer. Acquisition parameters were as follows: pulse width, 4.2 µs; sweep width, 18 kHz; acquisition time, 0.50 s; relaxation delay, 0.50 s; time domain, 18K; size, 32K; and digital resolution, 0.55 Hz. For all kinetic runs, a control spectrum was obtained (t ) 0 min) prior to adding the enzyme. The reaction was initiated by adding 50 ng to 0.5 mg of PI-PLC depending on enzyme activity. The volume of enzyme solution that was added varied from 0.25 to 10% of the sample volume (from
Novel Mechanism of Phosphodiester Cleavage 1 to 40 µL). The enzyme solutions were freshly prepared by dissolving the solid enzyme in distilled water. The concentration of the enzyme was determined spectrophotometrically ( ) 18 300 M-1 cm-1 at 280 nm). The amount of enzyme used for reaction was sufficient to achieve rates of about 0.05 µmol/min (1.25% conversion/min). Reactions were typically followed for about 1 h; however, reactions with a Vmax of
